Cancer Diagnosis Program | Biorepositories and Biospecimen Research Branch

Search BRD Papers

A critical spotlight on the paradigms of FFPE-DNA sequencing.

Author(s): Steiert TA, Parra G, Gut M, Arnold N, Trotta JR, Tonda R, Moussy A, Gerber Z, Abuja PM, Zatloukal K, Röcken C, Folseraas T, Grimsrud MM, Vogel A, Goeppert B, Roessler S, Hinz S, Schafmayer C, Rosenstiel P, Deleuze JF, Gut IG, Franke A, Forster M

Publication: Nucleic Acids Res, 2023, Vol. , Page

PubMed ID: 37351572 PubMed Review Paper? No

Purpose of Paper

Conclusion of Paper

Studies

1. Study Purpose

   This review summarizes recommendations for next generation sequencing of DNA isolated from formalin-fixed paraffin-embedded (FFPE) specimens. The authors also compare DNA quality metrics and sequencing results obtained when 50 and 200 ng of DNA isolated from a 13 year old matched frozen and FFPE liver specimen were used as input. Effects associated with different DNA repair approaches that included a commercial kit,an in-house in vitro sequential base exclusion repair protocol (IQBErepair), and bioinformatic processing, were also examined. This study used a paired set of a differentially preserved non-cancerous liver specimen (FFPE and frozen samples) that was stored for 13 years. The FFPE specimen was fixed in neutral buffered formalin and stored at room temperature, but no other details of specimen processing were included. FFPE slides were deparaffinized using Roti-histol before DNA extraction with the RecoverAll Total Nucleic Acid Isolation Kit for FFPE. DNA from the FFPE specimen was repaired using the in IQBErepair and the NEBNext FFPE DNA Repair Kit. The in-house IQBErepair protocol included modified nucleobase excision, polydeoxyribose incision, DNA repair preparation, DNA repair, and final nick sealing steps. DNA was extracted from the frozen specimen using the MagAttract HMW DNA Kit. DNA was quantified using the Qubit dsDNA HS Assay Kit. DNA fragment size analysis was conducted with a TapeStation electrophoretic system and DNA amplifiability was assessed with the KAPA hgDNA Quantification QC Kit. Sequencing libraries were prepared with repaired and unrepaired DNA using the Illumina Nextera Flex for Enrichment, Illumina Nextera DNA Flex, and Roche KAPA Hyper Prep Kits and hybridization capture was performed using the xGen Pan-Cancer panel v1.5 (target region size 801,904 bp) or the SeqCap EZ Prime Choice Probes and the SeqCap EZ Human Oncology Panel (target region size 2,919,578 bp). Libraries were sequenced on Illumina NextSeq (2x75 bp), HiSeq4000 (2x100 bp), and NovaSeq 6000 (2x150) machines.

   Summary of Findings:

   Based on the literature, the authors recommend fixing tissue in low-concentration (4% formaldehyde) neutral buffered formalin, storing FFPE blocks for <8 years, avoiding analysis of air-exposed FFPE sections, deparaffinization without agitation, overnight proteinase K digestion in an  solution, lyophilization rather than heating during concentration of the SNA sample, and avoiding unnecessary freeze-thaw cycling. The authors also recommend a DNA input ≥ 50 ng, a DIN >2.05, a ratio of 129 bp to 41 bp amplicons >10% (preferably ≥40%), increased coverage relative to frozen specimens, and targeted coverage ≥80% of the genome for next generation sequencing of DNA isolated from FFPE tissue. The authors also recommendation optimizing the bioinformatic workflow by considering variant allele frequency, removing reads with low mapping quality, and applying appropriate filters.

   DNA isolated from the 13 year old FFPE specimen had an average fragment size of 1490 bp, a DIN of 2.0, and a ratio of 129 bp to 41 bp amplicons of 5%. When sequencing results obtained from this FFPE specimen and a matched frozen specimen were compared (results from multiple sequencing workflows combined), the most common artefacts were C>T/G>A (~30% of artefacts) followed by C>A/G>T, T>A/A>T, and T>C/A>G (each ~17.5%). When artefacts rates were compared to those obtained using frozen specimens, the FFPE specimen had 7-fold more C>T/G>A artefacts and 5.75-fold more delC/delG artefacts; each of the other artefact types increased by 1.75-4-fold. In regions with low coverage, the AAF frequently exceeded 10% in the FFPE specimen. When DNA isolated from FFPE specimens were sequenced, a 2-fold higher duplicate rate and a smaller average library insert size (~90 bp versus 175 bp) were observed relative to frozen specimens. Mean coverage was also much more variable along the chromosome when DNA isolated from FFPE rather than frozen specimens was used. When 50 ng of enzymatically prepared DNA from FFPE specimens was repaired using the IQBErepair method, on-target coverage increased 53% and 80% at the two centers, respectively, and coverage uniformity increased, although only small non-significant improvements were observed when the NEBrepair Kit was used for DNA repair. Further, DNA repair with the IQBErepair Kit decreased the AAF and the sequence duplicate ratio. The AAF for the majority of artefact types and the duplicate ratio were significantly higher when DNA from FFPE tissue was repaired with the NEBrepair Kit compared to the IQBErepair method. The authors state the biggest advantage of the IQBErepair method for DNA repair was increased coverage of low-coverage regions, which decreased the number of artefacts with high frequency. Importantly, the IQBErepair method did not improve metrics when used with a DNA input amount of 200 ng that was sheared by ultrasonication; similarly, the IQBErepair method did notaffect sequencing results obtained with DNA isolated from frozen specimens. Use of two and three libraries reduced artefacts by 94% and 98%, respectively, and better results were observed when libraries included two different repair methods or an unrepaired library. In comparing bioinformatic analysis workflows for both 50 and 200 ng of input DNA isolated from FFPE specimens, the authors found that a deduplication approach using the UMI information performed better (more aligned reads and a lower AAF), than simple deduplication, molecular consensus by single read families, and duplex consensus by combining read families representing the double-stranded DNA. However, the total number of artefacts remaining per 10,000 bases was lowest when the duplex consensus approach was employed.  A higher number of aligned reads and a lower AAF occurred with 200 ng of DNA was used for sequencing compared to when 50 ng DNA was used for each bioinformatic workflow.

   Biospecimens

   • Tissue - Liver

   Preservative Types

   • Formalin
   • Frozen

   Diagnoses:

   • Not specified

   Platform:

   Analyte	Technology Platform
   DNA	Automated electrophoresis/Bioanalyzer
   DNA	Real-time qPCR
   DNA	Next generation sequencing
   DNA	Fluorometry

   Pre-analytical Factors:

   Classification	Pre-analytical Factor	Value(s)
   Biospecimen Preservation	Type of fixation/preservation	Formalin (buffered) Frozen
   Next generation sequencing Specific	Data handling	Simple deduplication Deduplication approach using the UMI information Molecular consensus by single read families Duplex consensus by combining read families representing the double stranded DNA
   Next generation sequencing Specific	Template/input amount	50 ng 200 ng One library Two libraries Three libraries
   Next generation sequencing Specific	Template modification	IQBErepair No repair NEBrepair Kit Enzymatically prepared DNA Ultrasonication sheared DNA

   View more